Capital Investment Plan 2026 — In Thailand, Order Sequence Decides

When a capital investment plan stalls at head office, the number that gets challenged is almost always the payback period. But the formula is not what is broken. What is broken is the assumption placed in the denominator, and the sequence in which things are bought, when, and under which incentive scheme. Thailand in 2026 is a market where capacity utilization has sunk into the 57% range and the tax windows around automation investment have been swapped out. This article builds exactly one model case, a 2,800,000 baht main package with 840,000 baht of peripherals stacked on top, and opens up every line item to show how the same equipment can produce a payback period of either 3.37 years or 8.93 years.

The first thing to break in a capital investment plan is the volume assumption

The denominator of a payback calculation usually contains “additional production volume.” That is the first thing to break.

What a utilization rate in the 57% range actually means

Thailand’s Manufacturing Production Index (MPI) for the second quarter of 2026 was -1.79% year on year. Average capacity utilization over the same period was 57.47%. On a single month basis, the MPI for June 2026 was -3.10% year on year, the steepest decline since November of the previous year.

Translated into the language of investment planning, two numbers appearing at the same time say this. More than 40% of installed capacity is idle. And orders are below last year.

Write “we will install robots to raise production capacity by 1.4 times and recover the investment from the gross margin on the additional volume” in that environment, and the approval request stops at a single question. “Who is going to buy that 1.4 times?”

Being unable to answer is not a sign that the author was unprepared. It is because equipment capacity and order intake are separate variables. If you raise capacity and orders do not follow, the only thing that grows is the denominator of the utilization rate. In fact, a utilization figure of 57.47% means that a little over 40% of the capacity that already exists is not being converted into revenue. It is entirely possible to read the plan as adding equipment on top of that and simply increasing the amount of unconverted capacity.

To be clear, this is not an argument against automating. It is an argument for not placing the basis of your payback on a variable you cannot move. Order volume is not something the plant controls. The moment you put something uncontrollable in the denominator, the payback period becomes a number that anyone can argue against.

Base it on unit cost times frequency, not on added volume

So what goes in the denominator instead. The answer is the unit cost times the frequency of what is actually happening today.

• Annual labor cost per person × number of people that can be removed
• Unit cost of one overtime hour × overtime hours that can be cut
• Cost of one rework × number of reworks that can be avoided
• Loss per scrapped part × number of scrapped parts that can be avoided

These four have something in common. They all hold at the current production volume. Even if orders do not grow, the fact that they did not grow does not reduce the effect. If orders fall, the effect does fall as well, but the sensitivity is far smaller than under an added volume assumption, and it can be verified against your own historical data.

A plan built on added volume is evaluated, from the moment it is approved, on whether the added volume materialized. A plan built on unit cost times frequency is evaluated on whether the unit costs and frequencies were as assumed. The latter is a matter that closes inside the plant, and it is a promise the person submitting the request can keep.

Do not delete the capacity gain, move it outside the payback calculation

That said, the fact that capacity increases is real. There is no need to erase it entirely. The practical approach is to leave it out of the payback calculation and record it separately as a qualitative side benefit.

Wording along these lines works well. “Payback for this investment is calculated using only the four items of labor cost, overtime, rework and scrap. The capacity gain is not included in the payback basis. When order intake recovers, the plant will have headroom to respond without further investment.” That avoids double counting and still gives you something to say if demand turns up.

What you must not do is derive a payback period from labor cost reduction and then add the gross margin from the added output generated by the freed machine time of the same equipment. The hours worked by the people removed and the freed machine time are two faces of the same single reality. Adding both means you have created two worlds that do not both exist. This is the failure mode seen over and over again.

Three questions that test the volume assumption

Open your own plan and check whether you can answer these three.

If you are looking at replacing aging equipment, the comparison is not “new equipment versus the status quo” but “new equipment versus continuing to repair what you have.” That framing is covered separately in when to decide on replacing aging equipment.

In Thailand, the tax windows changed in 2026

The second pillar. Anyone who wrote an automation investment request in Thailand up to 2025 and reuses the same material in 2026 will be stating something factually wrong. The scheme that used to carry these cases has expired, and schemes with a different character have taken its place.

Do not build on a scheme that ended at the end of 2025

The 200% deduction for machinery and software for automation systems, meaning the normal 100% plus an additional 100%, applied for the period from January 1, 2021 to December 31, 2025. The requirements were that the asset be unused, that it be installed in Thailand, that the automation project hold the certification required by the Revenue Department, and that it not be combined with other tax privileges.

In other words, this scheme cannot be built into a 2026 capital investment plan as a source of funds. Any 2024 or 2025 approval template still sitting on your internal drive very likely has this 200% deduction written into its assumptions. Stripping that out is the first job of 2026.

If you copy an old document and only swap the numbers, you will submit it without noticing that the scheme has lapsed. Being corrected by finance after approval costs the whole plan its credibility.

The windows open in 2026 and the closed ones in one table

Laying out the frameworks relevant to automation, energy efficiency and digital investment as of 2026 gives the following picture. What matters is that the eligible sectors, deadlines and caps are completely different in each case.

BOI Notification No. 4/2569 was published in the Royal Gazette dated March 31, 2026, and exempts corporate income tax for 3 years up to 50% of the investment for automation and robot adoption. The minimum investment of 1 million baht is roughly 4,500,000 yen at a fixed rate of 1 baht = 4.5 yen.

There is one limitation you cannot afford to drop. The categories covered by this notification are general automotive manufacturing (category 3.6) and PHEV/HEV manufacturing (category 3.8). Which means sectors outside automotive are not covered by this notification.

A line like “automation investment in Thailand is now tax exempt for three years” tends to circulate in internal chat, but it is a summary with the sector limitation removed. Electronics, food, metalworking, none of them can use the notification unless they fall under an automotive category. Leaving that limitation out of an approval request will be picked up in finance review without fail.

Note also that the scheme details written here are a summary of published information, and whether your own company qualifies depends on sector classification, registration status and the nature of the spending. Please confirm eligibility individually with a specialist in every case. The role of this article ends at handing over the list of points that need to be confirmed.

This is why the order of purchase changes what you keep

The fact that the schemes are fragmented means that the same purchase leaves you with a different amount depending on when and in what order it is ordered. There are three places where sequence matters.

First, whether spending crosses an accounting period. The 200% deduction for SME digital spending has a cap of 300,000 baht per accounting period. Because the unit of the cap is not simply an amount but an amount per accounting period, which period the spending lands in changes the result.

Second, whether schemes can be combined. The expired 200% deduction for automation systems carried a requirement that it not be combined with other tax privileges. Combination rules differ by scheme, so if you intend to run several schemes in parallel, you need to decide before ordering which spending is allocated to which scheme. Working out the allocation after the orders are placed leaves you with fewer combinations to choose from.

Third, whether certification and registration come before or after the order. The 150% deduction for energy efficient machinery applies to machinery certified with the DEDE five star label, and the 200% deduction for SME digital spending applies to DEPA registered services. The BOI notification involves an application procedure. Issuing a purchase order before checking the certification, registration and application requirements creates the risk of buying a model number that is not eligible. That is the most painful form of “the sequence decides.”

Splitting across accounting periods alone moves the deduction by 200,000 baht

Putting numbers on the first point. What follows is a model assumption for this article. It is not the actual result of any real company.

Assume that digital related spending such as production data collection and equipment data visualization comes to 500,000 baht, and that the full amount meets the eligibility requirements for the 200% deduction for SME digital spending. In practice an eligibility assessment is needed.

With a single order, only up to the cap of 300,000 baht is eligible for the 200% treatment, and the excess 200,000 baht stops at the normal 100%. The deduction is 300,000 × 2 + 200,000 = 800,000 baht.

Split across two periods, the 250,000 baht spent in each period is within the cap, so the whole amount is eligible for the 200% treatment. That is 250,000 × 2 = 500,000 baht per period, or 1,000,000 baht across the two periods.

The difference is 200,000 baht of additional deduction, roughly 900,000 yen of deductible expense at 1 baht = 4.5 yen. What you actually keep is that 200,000 baht multiplied by your own effective tax rate. Because rates differ by company, no cash figure is fixed here and the result is presented as an additional deduction amount. Apply your own effective tax rate to evaluate it.

But do not delay commissioning for the sake of the tax treatment

A check in the opposite direction is needed here. Splitting across two periods can mean that the functions tied to the later spending come on line six months to a year later.

Choosing the right comparison matters. What you compare is the benefit tied to the spending that was split against the cash kept from the additional deduction, that is the extra deduction multiplied by your effective tax rate. Do not compare it against the annual benefit of the whole project. Deferring only the digital spending does not delay commissioning of the machinery itself. Conversely, if the deferred spending covers functions tied directly to shop floor operation, such as production data collection or visualization, only the benefit generated by those functions is lost for six months.

The cash kept from the 200,000 baht of additional deduction is the amount after applying the effective tax rate. Whether the benefit lost through splitting exceeds that amount is the question. Run that one line of comparison for each spending item.

Put another way, splitting for tax reasons pays off only when splitting does not change the commissioning date. Spending such as software licenses or annual contracts, which does not affect operations even when it straddles periods, is a good candidate for splitting. Deferring construction work that requires a line stoppage is usually a loss.

The thing being optimized in the sequence is not the tax treatment alone. It is both the tax treatment and the commissioning schedule. Look at only one, and you will lose on the other every time.

The eight items missing from the quotation that move the investment by 20 to 30%

The third pillar. Just as fragile as how the assumptions are set is the investment amount itself.

Write the approval request from the equipment price alone and the rest shows up later

Quotations from vendors typically list the price of the main equipment. Robots, controllers, grippers, control panels. Put that total straight into an approval request and additions will appear after approval without fail.

The breakdown below follows the model assumptions of this article. The model is a Japanese owned electronic component processing plant in Rayong province, Thailand, with 210 employees on a two shift operation, in a sector other than automotive, carrying out automated feeding for two machining cells plus automation of an inspection process. The amounts are values set for the purposes of this article and are not the results of any real project.

The subtotal of the eight items is 95,000 + 140,000 + 60,000 + 175,000 + 165,000 + 55,000 + 105,000 + 45,000 = 840,000 baht. That is 30.0% of the 2,800,000 baht main package and 23.1% of the 3,640,000 baht total investment.

Converted at the fixed rate of 1 baht = 4.5 yen, the main package is roughly 12,600,000 yen, the eight items roughly 3,780,000 yen and the total roughly 16,380,000 yen.

Anyone who wrote the approval request from the equipment price alone has applied for an investment 30% too small. And because the payback period is proportional to the investment amount, their payback period looked 30% shorter as well.

The eight items get missed because they sit in nobody’s budget

The reason these eight items are missed is simple. They are ordered from different suppliers.

The robot comes from the robot maker, the electrical work from an electrical contractor, air from a piping contractor, safety fencing from another supplier, floor repair from the building maintenance department, and training from your own labor hours. Only the robot maker issues a quotation, so nobody else produces one.

The countermeasure is not to collect quotations supplier by supplier, but to pick the items up with a checklist ordered by the sequence of the work itself. Will the floor hold before the machine is brought in, is there enough power, is there enough air, is there still an operator walkway outside the fence. Work through it in that order and the question of who should issue which quotation resolves itself afterwards.

Show the contingency rather than hiding it

Even with the eight items stacked in, surprises still happen. The conservative scenario in this article adds a 10% contingency to the investment, giving 3,640,000 × 1.10 = 4,004,000 baht, roughly 18,020,000 yen.

Booking a contingency raises the investment and lengthens the payback period, which is exactly why people want to hide it. But a hidden contingency always surfaces later as a supplementary approval request. Supplementary requests erode trust regardless of the amount involved. Writing “a 10% contingency is included and will be returned if unused” from the start makes life easier once the plan is approved.

Do not present the payback period as a single number

The fourth pillar. Turning every assumption so far into numbers, the payback period is presented not as one figure but as three.

Fix the model assumptions and unit costs first

Everything below is a model assumption of this article. None of it is actual data from a real company. It is presented as a base that readers can substitute their own numbers into and recalculate.

The overtime rate is the only line without the 1.45 loading, because the premium is calculated on the base wage. Including the loading here would overstate the benefit, so the conservative treatment is used.

Only four benefit items, and only one world

Only the following four items are counted as benefits. All of them take the form of unit cost times frequency. The breakdown for the standard scenario is as follows.

696,000 + 36,000 + 121,800 + 171,600 = 1,025,400 baht, roughly 4,610,000 yen.

The headcount reduction breaks down as follows. Across the three target processes the staffing is 4 people per shift and 8 across two shifts, moving after implementation to 2 people per shift and 4 across two shifts. That is 8 − 4 = 4 people. The overtime reduction assumes 10 hours per month less for each of the remaining 4 people, so 10 hours × 4 people × 12 months = 480 hours.

On the other side, here is what was deliberately left out of the benefits. This is the gate that prevents double counting.

Leaving tax effects out of the payback period is deliberate. Putting an amount whose eligibility is not confirmed into the denominator means the payback period moves the moment eligibility fails. Writing it in the order of “if the scheme applies, payback comes sooner” makes the approval request safer.

Subtract the costs that rise after implementation

Automating also raises costs. The standard scenario books 225,000 baht per year, roughly 1,010,000 yen.

90,000 + 45,000 + 60,000 + 30,000 = 225,000 baht.

The annual net benefit is 1,025,400 − 225,000 = 800,400 baht, roughly 3,600,000 yen. The simple payback period is 3,640,000 ÷ 800,400 = 4.55 years.

Line up conservative, standard and aggressive assumptions

Same equipment, same formula, three different sets of assumptions.

The benefits and payback periods come out as follows.

Checking the arithmetic. Conservative is 522,000 + 18,000 + 84,000 + 109,200 = 733,200, then 733,200 − 285,000 = 448,200, and 4,004,000 ÷ 448,200 = 8.93 years. Aggressive is 870,000 + 45,000 + 142,800 + 218,400 = 1,276,200, then 1,276,200 − 195,000 = 1,081,200, and 3,640,000 ÷ 1,081,200 = 3.37 years.

For the same equipment, the payback period moves from 3.37 years to 8.93 years, a spread of about 2.65 times. That spread is the substance of the claim made at the top of this article, that the formula is not what is broken.

Note that the investment is not reduced even in the aggressive scenario. Benefits swing with the assumptions, but there is no reason anywhere to be optimistic about the investment amount. Lowering the investment in the aggressive scenario turns it from a range of assumptions into a wish.

Which assumption hurts most when it breaks

Starting from the standard scenario, here is the payback period when just one assumption is returned to its conservative value.

Headcount reduction dominates by a wide margin. One person of difference moves the payback by 1.26 years. Put the other way round, as long as the basis for the headcount reduction is solid, the plan survives a fair amount of drift in the other items.

There is a property here that is easy to overlook. Adding the six differences together gives 1.26 + 0.45 + 0.38 + 0.37 + 0.22 + 0.10 = 2.78 years, but the difference in the conservative scenario, where all six move at once, is 8.93 − 4.55 = 4.38 years. The addition falls short by 1.60 years.

The reason is that the payback period is a division, investment divided by net benefit. The smaller the denominator gets, the more a decline of the same size bites. Running sensitivity one item at a time and concluding “this looks fine” understates the case where several move together. Always produce the value for the case where everything moves at once as well.

How to firm up the basis for the headcount reduction

Because it is the assumption with the most leverage, this is the one to nail down with measurement.

1. Record the work in the target processes by shift and by operator, in time, for just one week
2. Sort the records into four categories, attending the machine while it runs, actual load and unload work, setup and checking, and other
3. Treat the robot as replacing, in principle, only the actual load and unload work
4. Where attending time is long, judge whether that person can genuinely be removed or will simply move to another process
5. If the answer is that they simply move to another process, do not count them in the headcount reduction

Skip steps 4 and 5 and the headcount reduction will always be overstated. What often happens in Thai plants is that nobody leaves and only the assignments change. Changing assignments does not reduce labor cost. Build a “before” and “after” staffing chart and confirm that the head count genuinely falls before putting it into the numbers.

There is also the design where only the night shift is unmanned while daytime staffing is unchanged. In that case what is reduced is the night shift allowance and the night shift headcount, and the structure of the calculation changes. Investment thinking that assumes unattended night operation is covered in the conditions that make unattended night operation work.

With interest at 1%, own funds or lease

Once the investment and the benefits are fixed, the next question is how to fund it.

Start by checking the interest rate level

The Bank of Thailand policy rate is 1.00%. It was cut by 0.25% to 1.00% at the February 2026 meeting and left unchanged at the June 2026 meeting.

The policy rate is not the cost of funds itself, but it does indicate an environment in which the time value of money is low. At this level, neither the opportunity cost of tying up own funds nor the interest burden on borrowing is likely to be large enough on its own to overturn an investment decision.

This article uses the simple payback period to keep the comparison straightforward. At a rate of 1.00%, introducing discounting is unlikely to change the ranking of payback periods, and the 2.65 times spread created by the assumptions is far larger than the effect of discounting. Running a precise present value calculation without first narrowing the spread of assumptions does not improve accuracy.

Put own funds and lease on the same footing

The following are also model assumptions of this article. Lease rates vary with each company’s credit standing, the machine type and residual value settings, so replace them with your own quotation.

• The investment is fixed at the standard scenario figure of 3,640,000 baht, and only the benefit side is varied across the three assumption sets in this section
• The lease runs 5 years, with total payments assumed at 1.12 times the principal
• The opportunity cost of own funds is calculated as an indicative figure by applying the policy rate of 1.00% directly

The annual uplift of 87,360 baht is 2.4 times the opportunity cost of 36,400 baht, since 87,360 ÷ 36,400 = 2.4. Note that the outstanding lease principal declines year by year, whereas the opportunity cost here is applied to the full principal. Both are rough indicators based on the full principal and neither is a strict effective annual rate.

At an interest rate of 1%, the uplift built into the lease rate is clearly larger than the opportunity cost of tying up own funds. On amounts alone, the conclusion favors own funds.

Cash flow, however, changes the conclusion

Do not decide on the amounts alone. A lease fixes the outflow each year, so it hurts in a year when the benefits do not materialize. For each of the three assumption sets, look at the annual net benefit less the annual lease amount.

Even in the standard scenario, cash during the lease term is close to breakeven, at 14,960 baht negative per year. In the conservative scenario, 367,160 baht per year has to be funded from other profits for five years.

In other words, a five year lease only breaks even once the standard scenario holds. A delay in benefits appearing, or a headcount reduction that comes up one person short, is enough to lock in a cash drain.

Note that lease accounting and the internal approval category depend on head office rules. The motive of “leasing because we want to avoid capitalization” is a separate matter from which option is financially better, so keep the two apart in the write up.

Four patterns of rejected approval requests and how to fix them on the spot

An automation investment in Thailand almost never gets approved on the grounds that everyone else is doing it. For reference, IFR World Robotics 2025 reports 542,000 units of new industrial robot installations worldwide in 2024, of which Asia accounted for 74%, with Europe at 16% and the Americas at 9%. At the same time, robot density in Asia is 131 units per 10,000 manufacturing employees, less than half of Western Europe’s 267 units and around two thirds of North America’s 204 units.

Units are concentrated in Asia, but density is still low. Those two numbers can be read as evidence for installing and as evidence against it. Which is why unit counts do not make a case in an approval request. The only things that do are your own unit costs and frequencies.

With that in mind, here are the four patterns of rejection.

Fixing the volume assumption pattern

This is the most common pattern, and the correction procedure is fixed.

Open the benefit table and mark every row that scales with production volume. Calculate what share of the gross benefit the marked rows represent. If it is above 50%, the plan is being held hostage by order intake. Delete the marked rows and recalculate the payback period using only what is left. That number is the promise you can keep.

In the model used here, the items among the four that scale strictly with production volume are rework and scrap, which come to 121,800 + 171,600 = 293,400 baht against a gross benefit of 1,025,400 baht, a share of 28.6%. Labor cost and overtime reductions survive moderate swings in volume as long as the staffing is not put back.

Fixing the single value pattern

Building three assumption sets requires no new investigation. The formula stays the same and only the values of each assumption move up and down. For deciding the width, use the observed variation for items where internal data exists, and swing generously to the conservative side for items where it does not. That is enough.

On top of that, add one line stating what the payback becomes if a given assumption fails. What approvers really want to know is not the average but the damage if it goes wrong.

Fixing the under quoted pattern

Staring at a quotation will not produce the missing items. The reliable method is to stand on the shop floor and trace the route with your finger, starting from the delivery path. Walk from where the truck stops to where the line will be installed, checking the width of every door on the way, steps in the floor, overhead piping, the nearest distribution board and the air take off point. Thirty minutes of this fills in five of the eight items.

Fixing the scheme dependent pattern

A plan funded by a tax scheme falls apart the moment the interpretation of that scheme changes. It is safer to write the tax effect not as something that shortens the payback period but as upside on top of an unchanged payback period.

Changing only the wording changes the risk after approval substantially. Instead of “payback is 3.0 years because we will use scheme A,” write “we request approval on a payback of 4.55 years, and if eligibility for scheme A is confirmed, it will be shorter than that.” With the latter, the plan survives even if the scheme falls away.

When you ask outside help for capital investment planning support, what can you actually ask for

The work described so far is voluminous and it crosses departments. Production engineering alone cannot assemble the historical data, and accounting alone does not know the frequencies on the floor. That is why capital investment planning support from outside comes into the picture.

The problem is that signing a contract while it remains vague what can and cannot be delegated leads to disappointment.

Look at the right hand column. Most of the variables that determine the payback period exist only inside your own company. What can be delegated is the template and the mechanism, not the numbers themselves.

The first thing to ask for is how to collect data, not how to calculate

In most cases the first bottleneck is not calculation capability. It is that the frequency in unit cost times frequency is unknown.

How many reworks occur per month, how many changeovers happen per month, how many minutes each one takes. Without these in a log, the basis for the benefits becomes the shop floor’s sense of things. Numbers built on a sense of things do not survive a sensitivity analysis.

So the first place capital investment planning support creates value is getting to a state where what happens and how often it happens in the target process is recorded automatically. With the data in hand, the next investment decision does not require repeating the same work.

TOMAS TECH CO., LTD., based in Bangkok, supports plant IT and OT adoption for Japanese manufacturers across Thailand and ASEAN, centered on the PEGASUS production management and energy management system. In the context of capital investment planning, we are often consulted not about whether the investment itself is justified but about the step before that, whether the frequency, time and defect data needed for that judgment is being captured at all.

If you are starting the automation review from carving out the right process, the approach is set out in what to do first in automation consulting.

What to have ready before you ask

Having these four items ready before approaching an outside partner makes the first conversation concrete.

1. Production records for the target process for the last 12 months, monthly is fine
2. The staffing chart for the target process, by shift, headcount and job type
3. Rework and scrap counts for the last 12 months, to the extent they are known
4. The head office approval format and the payback period level it expects

The fourth is especially important. Calculating without knowing the payback level head office requires means doing the work twice. If the required level is three years, a project whose conservative scenario comes out at 8.93 years needs to be revisited starting from how the investment scope is drawn.

Frequently asked questions

What is a reasonable payback period for automation

There is no single right answer. The level head office requires differs by company, and it also varies with the useful life of the equipment. What matters in practice is to present a range rather than a single number, and to be able to explain what drives the range. In the model case used here, the payback period for the same equipment moved from 3.37 years to 8.93 years, a spread of about 2.65 times. Disclosing that spread up front shifts the discussion from “why is it 4.55 years” to “which assumptions do we have to hold in order to get 4.55 years.” When you present a payback period from the plant side, always present the assumption with the most leverage alongside it.

How should the cost benefit of a robot installation be calculated

Break the benefits into unit cost times frequency and count only the items that hold at the current production volume. This article used four items, direct worker reduction, overtime reduction, rework reduction and scrap reduction, and deliberately excluded added output. The reasons are that with average utilization at 57.47% there is no guarantee freed machine time turns into revenue, and that counting both the hours of the people removed and the freed machine time would count the same reality twice. On the investment side, add to the equipment price the eight items of installation, power, air, frames, safety fencing, delivery, commissioning and training. In the model here, those eight items came to 30.0% of the equipment price and 23.1% of the total investment.

Is it forbidden to mention added output in a capital investment approval request

It is not forbidden to mention it. It is forbidden to put it in the denominator of the payback period. The capacity gain is real, so recording it separately as a side benefit is useful. Kept separate, it avoids double counting, and it also lets you explain that recovering demand can be met without further investment. The moment it is mixed into the payback calculation, the whole plan stops at the question of who is going to buy the extra output.

Are there tax schemes available for automation investment in Thailand in 2026

To start with, the 200% deduction for machinery and software for automation systems applied from January 1, 2021 to December 31, 2025 and has already ended. It cannot be built into a 2026 plan as a source of funds.

Among the current frameworks there is BOI Notification No. 4/2569, which exempts corporate income tax for 3 years up to 50% of the investment for automation and robot adoption, with a minimum investment of 1 million baht and an application deadline at the end of 2027, but it is limited to general automotive manufacturing (category 3.6) and PHEV/HEV manufacturing (category 3.8). Sectors outside automotive are not covered. Separately there is the 150% deduction for energy efficient machinery under Royal Decree No. 805, effective from March 3, 2026 to December 31, 2028 and applying to machinery certified with the DEDE five star label, and the 200% deduction for SME digital spending, from June 24, 2025 to December 31, 2027, capped at 300,000 baht per accounting period and applying to DEPA registered services. Eligibility varies with sector classification, registration status and the nature of the spending, so please confirm your own case with a specialist.

If we ask for outside capital investment planning support, where does it start

In most cases it starts not with calculation but with taking stock of the data. As long as benefits are expressed as unit cost times frequency, an unknown frequency cannot enter the calculation. The number of reworks, the number of changeovers and how long each takes, downtime by cause. Building an estimate while none of that is in a log produces numbers that cannot survive a sensitivity analysis. What can be asked of an outside partner is the template for breaking down benefits, the identification of items missing from quotations, the design of three assumption sets, an organized view of the schemes, and building the mechanism for capturing operating data continuously. The measured data and the decision to genuinely remove people can only be produced inside your own company.

Summary

What breaks in a capital investment plan is not the payback formula. It is how the assumptions are set and the sequence in which things are ordered.

First, the volume assumption. Thailand’s MPI for the second quarter of 2026 was -1.79% year on year, average capacity utilization was 57.47%, and the June 2026 MPI was -3.10% year on year. Placing added output at the base of the payback in that environment stops at the question of who is going to buy it. Base it instead on unit cost times frequency. Labor cost, overtime, rework, scrap. Count only the items that hold at the current production volume.

Second, the tax windows have changed. The 200% deduction for automation systems ended on December 31, 2025. BOI Notification No. 4/2569 exempts corporate income tax for 3 years up to 50% of the investment, but it is limited to general automotive manufacturing (category 3.6) and PHEV/HEV manufacturing (category 3.8). The 150% deduction for energy efficient machinery and the 200% deduction for SME digital spending each have different scope, deadlines and caps. That is why the sequence of what is bought, when, and under which scheme matters. Under the model assumptions of this article, 500,000 baht of digital spending concentrated in one accounting period versus split across two periods gave deductions of 800,000 baht and 1,000,000 baht respectively, a difference of 200,000 baht. But if splitting delays commissioning, you need to set the benefit tied to that spending that is lost against the cash kept from applying your effective tax rate to the 200,000 baht. That is the reason not to optimize for the tax treatment alone.

Third, the granularity of the quotation. Against a 2,800,000 baht main package, the eight items of installation, power, air, frames, safety fencing, delivery, commissioning and training come to 840,000 baht. That is 30.0% of the main package and 23.1% of the 3,640,000 baht total investment. Anyone who wrote the approval request from the equipment price alone has applied for an investment 30% too small.

Fourth, present the payback period as a range. In the model here, conservative is 8.93 years, standard 4.55 years and aggressive 3.37 years, a spread of about 2.65 times. The assumption with the most leverage is the headcount reduction, where one person moves the result by 1.26 years. And adding the six sensitivities individually gives 2.78 years, while moving them together gives 4.38 years. Because it is a division, adding the individual effects understates the damage.

On funding, at a policy rate of 1.00%, the uplift on a five year lease at 87,360 baht per year came to 2.4 times the opportunity cost of own funds at 36,400 baht per year. Even so, in the standard scenario the annual lease amount of 815,360 baht against a net benefit of 800,400 baht leaves cash close to breakeven during the term. Judge the financial advantage and the cash flow safety separately.

The first thing to do is neither vendor selection nor model selection. It is to measure, for one week only, what happens and how often in the target process. Until that gap is filled, no formula will turn the numbers into anything more than a sense of things.

Capital investment in Thailand moves on several fronts at once, the incentive schemes, the labor situation and the level head office demands. TOMAS TECH is based in Bangkok and supports Japanese manufacturers in plant IT and OT/FA, and we are often consulted not about whether an investment is justified but about the step before it, whether the data needed for that judgment is being captured. Where to begin the measurement, which of the eight items matter most at your site, how wide to swing the three assumption sets, we are happy to work through these together in light of your site’s situation. You are welcome to get in touch through the contact page.

References

• Tilleke & Gibbins, Thailand Unveils New Incentives for Automotive and HEV/PHEV Manufacturing — https://www.tilleke.com/insights/thailand-unveils-new-incentives-for-automotive-and-hev-phev-manufacturing/3/
• Sherrings, Automation Systems Costs in Thailand — https://sherrings.com/automation-systems-costs-thailand.html
• Bizwings, Thailand Introduces Tax Incentives for Solar Rooftop Installation and Energy Efficient Machinery — https://www.bizwings.co/post/thailand-introduces-tax-incentives-for-solar-rooftop-installation-and-energy-efficient-machinery
• LexNova Partners, Thailand SME Policy Updates — https://lexnovapartners.com/thailand-sme-policy-updates/
• Xinhua, Thailand manufacturing production index and capacity utilization — https://english.news.cn/20260727/f8b7ceb202d04b1f8facbcbf44ed60ce/c.html
• Trading Economics, Thailand Interest Rate — https://tradingeconomics.com/thailand/interest-rate
• JETRO, Minimum wage in Bangkok set at 400 baht per day from July 1, 2025 — https://www.jetro.go.jp/biznews/2025/07/b21007a1ac8f7fca.html
• IFR, Global Robot Demand in Factories Doubles over 10 Years — https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years